AI now lets IT teams write a script or build a device configuration in seconds. Undoing a bad one still takes most teams hours, and often days.
That gap is the central finding of Fleet Device Management’s Apple in the Enterprise report, released Oct. 7. Fleet surveyed more than 250 enterprise IT practitioners who manage Apple devices. The respondents work inside enterprises, not MSPs, but the pattern applies to any provider pushing changes to client devices.
AI-written changes are already in production
According to the report, 86% of respondents allow AI-written output to reach production devices after some form of review. In the past 30 days:
- 61% used AI to author an MDM profile or configuration
- 62% used AI to write scripts and code
- 68% used AI for troubleshooting or log analysis
AI assistants are now part of the daily toolkit. Fifty-seven percent of respondents regularly use LLM assistants, more than the 53% who regularly use MDM consoles. Forty-six percent use a code editor and 42% work with CI/CD runners.
Recovery is the weak link
Nearly 70% of respondents can’t roll back a bad configuration change within an hour, and 43% need more than a day. Another 26% can recover the same day, but only with manual intervention.
“AI can help push changes out in seconds, while a mistake can drain hours of valuable IT time and resources,” said Allen Houchins, Fleet’s CIO. “Teams need a clear record of what changed and a fast way to undo it when something goes wrong. Speed is only an advantage if the pace of recovery matches.”
Only 28% of respondents use both human review and version control for AI-generated changes, the combination that gives them a record of every change and a path back. Fleet CEO and co-founder Mike McNeil said Mac admins have effectively become engineers without engineering’s safety net:
Why this hits MSPs harder
An internal IT team that pushes a bad profile breaks one company’s devices. An MSP that pushes the same change across its client base can break dozens, and each hour of recovery lands on the service desk and the SLA.
The report also shows how mixed most environments are. Only 8% of respondents manage Apple devices exclusively. Apple adds its own pressure: 78% are using or piloting declarative device management (DDM), and 31% say Apple’s release cadence is the hardest part of Mac administration.
A change-control checklist for AI-written configurations
The fix isn’t avoiding AI. It’s treating device configuration the way developers treat code. Before an AI-written script or profile reaches a client device:
- Store it in version control. Every script and profile should live in a repository, so you can see what changed, who approved it and what to revert to.
- Name a human reviewer. “Some review” isn’t a policy. Decide who signs off on AI-generated changes, and what they check.
- Write the rollback first. If you can’t describe how to undo a change, it isn’t ready to deploy.
- Stage the rollout. Push to an internal test group, then one pilot client, before the full client base.
- Time a rollback drill. Measure how long recovery actually takes, and close the gap before a real incident does it for you.
The full Apple in the Enterprise report is available from Fleet.
